As SOA and Web services adoption in the industry is gaining more momentum, the need to get quick wins and to show the value of adopting new (or old) paradigms is weighed against the risk of facing the repercussions of slapping something together in a quick and dirty fashion and paying the higher cost later. Many of our smart clients (not to be confused with .NET smart clients) are putting together the right groups to facilitate the adoption of these new technologies across their organizations.

The deployment of SOA is centered on governance. In order to have an efficient governance process, central groups that can act as COEs (Centers of Excellence) need to be in place before portfolios are allowed to develop and deploy functionality for their respective applications. There are too many technologies, business processes, vendor products, and confusing messages in the mix for any organization to leverage the benefits of adopting SOA through Web services effectively. Research needs to be conducted, products need to be evaluated, eliminated, and selected, and patterns and guidelines need to be published to address the common needs of the applications. This obviously has to be balanced against the drivers from business to show value ASAP.

Obviously cost is the main hindrance. It is very hard for a COE to show tangible cost savings. Unfortunately, the software industry hasn't developed appropriate metrics to assess the value provided by such groups. Fortunately, factors such as unfamiliarity with new technology and lack of required skillsets in individual application groups drives the need for establishing centers of excellence. It is still a hard sell to maintain such bodies, which are viewed by business owners as cost centers.

SOA lends itself very well to the outsourcing and off shoring model. An interesting side effect of this development is that a center of excellence can be distributed across geographies to manage the economies of scale. There is no getting away from the fact that the ultimate distribution and sharing of knowledge needs to be at the location where the applications are being developed. But this does not mean that the work cannot be distributed.

While this is an attractive proposition, it requires careful planning to set it up. To set up an effective distributed COE, appropriate roles need to be defined and staffed at the client site, as well as at the offshore/nearshore locations. Effective project management and processes need to be set up to produce deliverables such as white papers, vendor evaluations, newsletters, FAQs, blueprints, architecture cookbooks, patterns and guidelines documents, etc. Appropriate infrastructure needs to be set up to enable effective communication and knowledge sharing.

As of yet, this is not a model that many organizations have thought through or established. Fortunately, in our client engagements, we have seen traction from several clients and genuine interest in taking advantage of the distributed nature of large global consulting shops. At Infosys we have been able to offer attractive value propositions for setting up such organizations. One of the primary requirements to be able to effectively deliver on promises made in such initiatives is to have a well-established research facility and alliances with the appropriate vendors in the SOA and Web services space. The distributed COE model that we have deployed is illustrated in Figure 1. The diagram illustrates how a center of excellence for Web services and SOA is established for a U.S. client, with portfolios distributed across different states. The states shown in the diagram have been changed from the actual sites where the COE has been implemented to maintain client confidentiality.

The bottom line is that if you have a large sized IT with multiple portfolios, and if you don't consider standardization and governance across your multitude of applications, then the cost that will be paid in the long run offsets the savings that you will achieve in the immediate future. Don't be penny-wise and pound-foolish.